Weakly hierarchical lasso based learning to rank in best answer prediction

摘要: In community question and answering sites, pairs of questions their high-quality answers (like best selected by askers) can be valuable knowledge available to others. However lots receive multiple but askers do not label either one as the accepted or even when some replies answer questions. To solve this problem, prediction has been important topics in social media. These user-generated ten consist "views", each capturing different (albeit related) information (e.g., expertise asker, length answer, etc.). Such views interact with other complex manners that should carry a lot for distinguishing potential from Little existing work exploited such interactions better prediction. explicitly model these information, we propose new learning-to-rank method, ranking support vector machine (RankSVM) weakly hierarchical lasso paper. The evaluation approach was done using data Stack Overflow. Experimental results demonstrate proposed superior performance compared approaches state-of-the-art.